Abstract

View Abstract PDF

This project redevelops a framework for communicating with the Kuri, primarily using rosbridge, Javascript and HTML5. In doing so, we investigated the methods for converting data from the Kuri’s multidimensional frame to the two-dimensional (2D) frame of our Graphical User Interface (GUI).

Numerical data was collected using rosbridge publishers and subscribers. In our rosbridge code, subscribers to the odometry and laser-scan topics of the robot were written in JSON format. The backend included a rosbridge publisher to the robot’s mobile-base topic, thus allowing the GUI to control its wheels.

Three-by-three transformation matrices were also constructed to evaluate the odometry and laser-scan ranges of the robot during motion. The matrices were multiplied by the position points of our 2D marker, to determine this figure‚Äôs new position on the screen once the actual robot moved. These incorporated quaternion-to-euler and cartesian-to-polar conversions to navigate between the dimensions of our figure and the robot. 

Preliminary results show that our GUI successfully allowed for user interaction with the robot. Whenever a joystick moved the robot, our marker also moved on the screen in the direction of the robot‚Äôs motion. This confirmed that our resultant matrices were tracking and recording the robot‚Äôs odometry, and that our unit conversions had appropriately scaled the movement of the robot to our canvas‚ and frame.

However, in testing the driving features of our GUI, there were occasional discrepancies between the rotation of the marker and that of the robot. We hypothesized that this was due to the range restrictions of the inverse-sine and inverse-cosine functions we used in evaluating the quaternions. Further research will address the discrepancy by formulating an alternative method for navigating between the frames of the robot and our 2D figure.